Self-Host AI with Ollama, OpenWebUI and Authelia

This project provides a Docker-based setup to self-host AI services using Ollama, OpenWebUI and Authelia (OIDC). Follow the instructions below to get started.

Prerequisites

• Docker
• Docker Compose
• NVIDIA GPU with drivers installed (for GPU acceleration)
• NVIDIA Container Toolkit (if necessary)

Setup

1. Clone the repository:

   git clone https://github.com/tdharris/ollama-openwebui.git
   cd ollama-openwebui

2. Copy the sample environment file and update it with your configuration:

   cp .env.sample .env

   Edit the .env file to set your specific configuration values.

3. Build and start the services:

   docker compose up -d

   This will start the Ollama and OpenWebUI services.

Configuration

Environment Variables

• OLLAMA_IMAGE_VERSION: Version of the Ollama image to use (default: latest).
• OLLAMA_PORT: Port on which Ollama will be accessible (default: 11434).
• OPEN_WEBUI_IMAGE_VERSION: Version of the OpenWebUI image to use (default: cuda).
• OPEN_WEBUI_PORT: Port on which OpenWebUI will be accessible (default: 3002).
• OAUTH_CLIENT_ID: OAuth client ID for OpenWebUI.
• OAUTH_CLIENT_SECRET: OAuth client secret for OpenWebUI.
• OPENID_PROVIDER_URL: URL of the OpenID provider.
• OAUTH_PROVIDER_NAME: Name of the OAuth provider.
• OPENID_REDIRECT_URI: Redirect URI for OpenID authentication.

Note: See Authelia OIDC Open WebUI for more information on how to configure Authelia with OpenWebUI.

Volumes

• ollama: Stores Ollama data.
• open-webui: Stores OpenWebUI data.

Usage

Accessing the Services

• Ollama: Accessible at http://localhost:${OLLAMA_PORT}
• OpenWebUI: Accessible at http://localhost:${OPEN_WEBUI_PORT}

Stopping the Services

To stop the services, run:

docker compose down

Updating the Services

To update the services to the latest version, pull the latest images and restart the services:

docker compose pull
docker compose up -d

Download Models

More models can be found on the Ollama library⁠.

To download a model, use the following command:

docker exec -it ollama ollama run llama3

Note: Replace llama3 with the model name.

Troubleshooting

Check the logs of the services if you encounter any issues:

docker compose logs -f

License

This project is licensed under the MIT License.